get ///jab/ must CLONE the submodule into cwd and INHERIT its tracked pin-branch; today it in-place-refreshes the source tree and records trunktrees example 1 (lines 155-166): jab get ///jab/ inside an empty work/TKT-123 clones the main tree's jab submodule there, and the fork inherits the pin-branch the source tracks (?/jab/.beagle, lines 193-197). Neither half holds. (a) authorityRepo (loop.js:189-217) consumes the //+empty-host authority, rescopes the run to the main tree's jab/ and blanks the arg — get's plain driver then in-place FF-refreshes THE SOURCE (inRepoSeed get.js:1031-1061, 429-434), cloning nothing; get.js:286-292 documents this as known ("a //name nav authority is the CONTEXT…, NOT a clone SOURCE" — the BE-030 "separate follow-up", unticketed until now). (b) Even on the working local-source path, seedLocal (get.js:170-189) records ?""#tip = TRUNK: rem.branch comes only from an explicit ?/proj/branch query (get.js:91-93); resolveLocalSource (get.js:141-165) never reads the source's attachedBranch. The mis-tracked fork poisons the whole example chain: example 4's commit then advances trunk, not ?/jab/.beagle.
///jab/ — empty authority is falsy (get.js:261-267); only schemed/.be-path sources route to handleSeed+seedDest (get.js:296-319)..be FILE row-0 store redirect — the store half works; only the branch is dropped.//ABC-123 checkout the rev worktree ABC-123 is on" — rev only, silent on tracked-branch inheritance (wiki audit omission 17).//X////sub/ OPERAND to get, when cwd is NOT inside X, is a clone SOURCE: seed into cwd (or the named cell), not an in-place refresh of X.attachedBranch (wtlog.open(src).attachedBranch() via branchlib) — the fork tracks the source's pin-branch.?/jab/.beagle#<tip>, then a commit there reads "ahead" of that branch.get at home.get ///jab/ → clone lands in cwd, tracked row = source's pin-branch?query override winning